Print

[TotoMMB]

http://blog.prosper.com/2010/03/16/lender-servicing-fee-refunds/

ETA: decided it was better to capture the contents of the blog entry.

Recently we received an email from one of our lenders calling attention to a very odd looking loan payment transaction.  The situation involved a loan from 2007 that had gone delinquent, but had been recently cured.  On loans of that vintage, Prosper forfeited servicing fees during the time that a loan was more than 30 days delinquent.  However, on the “post cure” payments, the servicing fee was being inaccurately equated to the entire payment amount.

In our initial investigation, we found 30 payments spread across 23 loans that were definitely impacted.  We have processed refunds on those loans during Monday night’s batch.  These have been posted to the lender’s account as “Service Fee Refund” with a transaction date around 1 am on 3/16.  We have identified another 45 payments in the $10 to $25 dollar range that we are reviewing to see if they are impacted.   That review should be completed by the 17th and the refunds for the impacts transactions issued that night.

Because of the module in which this logic resides, the testing and migration of a “fix” into the production environment is complicated.  As such, we are creating a “watchdog” query that will run on a weekly basis to catch and refund an subsequent errors until the fix is installed in production.

"So, what exactly would you say you do here...?"

A "watchdog" query? That's a new one to me. "We won't fix it, but we'll have something that tells us when something may be amiss". I can't wait to log onto incompetent.com to check my balances on April 1. (The domain appears to be available).

[Mark12547]

A "watchdog" query? That's a new one to me. "We won't fix it, but we'll have something that tells us when something may be amiss". I can't wait to log onto incompetent.com to check my balances on April 1. (The domain appears to be available).

We have a few of those at work. If you are aware of a known bug that would be too difficult to fix, but you have a way of finding out when that bug occurred and the data can be fixed after the fact, you periodically generate a report of when that occurred so you can fix it after the fact.

Yes, it is better if the error doesn't occur, but sometimes fixing the bug is too complicated or the bug resides in a vendor's product so you aren't allowed to touch it.